Aspects of the present disclosure are related to systems, apparatus, and methods for training a chatbot, such that a user can ask a question and the bot can provide the answer using natural conversation. The solution disclosed herein creates a conversation engine that can be trained automatically by feeding in articles, FAQs, or documents on specific topics. The conversation engine can provide answers to user's questions on the trained topic through a natural conversation. The conversation engine can provide web service endpoints for training the engine on specific topics, and which can also be used during conversation with the user on that topic. When users ask a question on a topic to the chatbot, the chatbot can invoke conversation service endpoints, together with the intents, and the conversation service can provide the response via the chatbot.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A computer-implemented method comprising: at a server having one or more processors and memory storing one or more programs for execution by the one or more processors so as to perform the method: communicating electronically with a chatbot service to receive a digital request from a user interface; identifying, by the chatbot service, an intent from the digital request; transmitting, by the chatbot service, the intent to a webhook for interacting with a conversation engine for conversing with a user; receiving, at the webhook, the intent from the chatbot service and orchestrating a communication with a backend microservices namespace to provide a response to the intent; and determining, by the backend microservices namespace, the intent as one of a chit chat or a question; and wherein the orchestrated communication with the backend microservices namespace comprises: invoking at least one AI-based microservice that performs real-time processing of the digital request by transforming the digital request into a vector representation using a transformer encoder; comparing the vector to a set of previously embedded vectors stored in a Facebook AI Similarity Search (FAISS) based similarity search database; and returning a contextually relevant response to the chatbot service for presentation to the user.
2. The method of claim 1, wherein the digital request is a voice message or a text transmitted using a web or mobile device interface.
3. The method of claim 1, further comprising: generating, by the backend microservices namespace, a response to the intent using a chit chat generation microservice if the intent is chit chat; and transmitting the response to the user.
4. The method of claim 1, further comprising: generating, by the backend microservices namespace, a response to the intent directly to the user if the intent is a question; and transmitting the response to the user; and wherein the response is generated from a repository of well documented responses to questions.
5. The method of claim 1, further comprising detecting, using a frequently asked question (FAQ) similarity microservice of the chatbot service, similarities between the intent and a previously asked and answered intent.
6. The method of claim 5, wherein the previously asked and answered request is learned content stored in a knowledge base accessible by the chatbot service.
7. The method of claim 5, wherein the previously asked and answered request is stored locally in cache to immediately provide the response to the user in the form of a text.
8. The method of claim 5, wherein the frequently asked question (FAQ) similarity microservice of the chatbot service identifies the response to the user based on a context of the intent.
9. The method of claim 1, wherein the conversation engine is trained automatically trained by sending articles, frequently asked questions (FAQs), or documents on specific topics to the conversation engine.
10. The method of claim 1, further comprising: converting the digital request, using a frequently asked question (FAQ) similarity microservice, into a vector by transmitting the digital request through a transformer encoder; and determining a most similar vector using an AI Similarity Search Database Microservice to find a most similar vector.
11. The method of claim 10, further comprising performing a vector space search for the encoded digital request to find an answer to the intent.
12. A computer-implemented method comprising: at a server having one or more processors and memory storing one or more programs for execution by the one or more processors so as to perform the method: communicating electronically with a chatbot service to receive a digital request from a user interface; identifying, by the chatbot service, an intent from the digital request; transmitting, by the chatbot service, the intent to a webhook for interacting with a conversation engine for conversing with a user; receiving, at the webhook, the intent from the chatbot service and orchestrating a communication with a backend microservices namespace to provide a response to the intent; determining, by the backend microservices namespace, the intent as one of a chit chat or a question; converting the digital request, using a frequently asked question (FAQ) similarity microservice, into a vector by transmitting the digital request through a transformer encoder; determining a most similar vector using an AI Similarity Search Database Microservice to find a most similar vector; performing a vector space search for the encoded digital request to find an answer to the intent; and wherein the vector space search and similarity matching are performed using a Facebook AI Similarity Search (FAISS) based similarity database and transformer-based encoder to identify and return a contextually relevant response in real-time.
13. The method of claim 12, wherein the digital request is a voice message or a text transmitted using a web or mobile device interface.
14. The method of claim 12, further comprising: generating, by the backend microservices namespace, a response to the intent using a chit chat generation microservice if the intent is chit chat; and transmitting the response to the user.
15. The method of claim 12, further comprising: generating, by the backend microservices namespace, a response to the intent directly to the user if the intent is a question; and transmitting the response to the user; and wherein the response is generated from a repository of well documented responses to questions.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
May 10, 2022
May 27, 2025
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.